The Minimum System Requirements of DataMiner require a CPU PassMark score of 10000. However, I have seen systems where this requirement was satisfied, yet still performance seemed a bit off.
In one particular case, we increased the poll speed of a group containing an html file that needed to be parsed through SLScripting, which prompted a CPU increase of 18% for SLScripting.
This made sense as the polling was an order of magnitude heavier than before.
When testing said system, we noticed that the CPU PassMark score indeed matched 10000, however, the Single Threaded CPU score barely broke 1000, which seems to be on the low end of the spectrum.
Is it correct that Single thread performance might have a big impact for the workings of certain DataMiner processes such as the parsing of incoming responses?
Is there a recommended Single Threaded CPU PassMark?
There is indeed no recommendation on the Single Thread Rating because DataMiner typically works multithreaded and therefore the overall CPU benchmark is more important as requirement. Single Thread Ratings also tend to have less variations as modern CPUs are typically adding more cores to increase the speed.
That being said, what you process in one single thread will indeed go faster or slower depending on that Single Thread Rating (at least to some extend). Therefore it might indeed influence specific cases just like many other specifications of the hardware, but overall this requirement is less important for DataMiner.
Also note that this requirement is typically more important for client machines and game PCs. And it’s not because you can get a certain rating on one single thread, that you can have it on all available threads simultaneous. E.g. when hyperthreading is being used, one thread can use a complete core if the other thread is not doing anything. This might be the case on a client/game PC, but in DataMiner or in virtual environments this might not be the case and you will never reach the Single Thread rating you can find on the internet. Therefore I believe 1k for a single thread is not a bad number and is actually a typical number for VMs. It even goes to 700-800 for some threads…